| Title: | Digital Fortran |
| Notice: | Read notes 1.* for important information |
| Moderator: | QUARK::LIONEL |
| Created: | Thu Jun 01 1995 |
| Last Modified: | Fri Jun 06 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 1333 |
| Total number of notes: | 6734 |
OpenVMS Alpha V6.2
Digital Fortran 90 V7.0-20, ok
Digital Fortran 90 V7.1-330, NONGRNACC with /GRANUARITY=BYTE
The following example will fail to compile after upgrading to
DFAV V7.1, if you use the default or /GRANUARITY=QUADWORD it
works. Nothing in the release notes about this.
c_send_buffer(9:102) = 'test'
........^
%F90-W-NONGRNACC, Unable to generate code for requested granularity
Regards, Drew Sanford
C970326-6830
C
C f90/granularity=byte err.for
C
character*(102) c_send_buffer
c_send_buffer(9:102) = 'test'
END
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 1255.1 | QUARK::LIONEL | Free advice is worth every cent | Fri Apr 11 1997 15:04 | 3 | |
One of our GEM buddies want to comment? Steve | |||||
| 1255.2 | GEMGRP::HAMNETT | Needs Must! | Fri Apr 11 1997 17:16 | 16 | |
The warning is indicating that the user had requested byte granularity while doing a string move. The Libots routines OTS$MOVE, OTS$FILL and OTS$ZERO on VMS are quadword granular. This is true for Libots V1.3-013 and all previous versions. In response to a bug report, where the customer was depending on longword granular behavior, (so the libots routines were writing over adjacent code) we added the warning message to the compiler. We also submitted V1.3-014 of Libots to VMS with longword versions of these routines. However, as far as I know, that has never been released. Further, we are planning to make the Libots routines Byte granular in the Raven release, at which point this message will be removed from the compiler. Lucy | |||||
| 1255.3 | byte-granular only when it's cheap? | WIBBIN::NOYCE | Pulling weeds, pickin' stones | Mon Apr 14 1997 09:05 | 7 |
> Further, we are planning to make the Libots routines Byte granular in the > Raven release, at which point this message will be removed from the compiler. Lucy, I thought the plan was to provide byte granularity only on processors that support the store-byte instruction, such as EV56, PCA56, EV6. Won't users with older processors still be exposed? -- Bill | |||||
| 1255.4 | GEMEVN::GROVE | Mon Apr 14 1997 09:36 | 4 | ||
Bill is correct - the new libots provides byte granularity only on EV56
(and newer) processors that have STB instructions.
/Rich
| |||||